• Proceedings of the KSPS conference
• /
• 2003.05a
• /
• pp.207-210
• /
• 2003

In this paper, the combinations of speech enhancement techniques are experimented. Specifically, the spectral subtraction, KLT based comb-filtering, and their combinations are applied to the Aurora2 database. The results show that recognition accuracy is improved when KLT based comb-filtering is applied after spectral subtraction.

• Journal of Korean Society for Geospatial Information Science
• /
• v.16 no.2
• /
• pp.49-56
• /
• 2008

Image registration is one of the critical techniques of image mosaic which has many applications such as generating panoramas, video monitoring, image rendering and reconstruction, etc. The fundamental tasks of image registration are point features extraction and tracking which take much computation time. KLT(Kanade-Lucas-Tomasi) feature tracker has proposed for extracting and tracking features through image sequences. The aim of this study is to demonstrate the usage of effective and robust KLT feature detector and tracker for an image registration using the sequence image frames captured by UAV video camera. In result, by using iterative implementation of the KLT tracker, the features extracted from the first frame of image sequences could be successfully tracked through all frames. The process of feature tracking in the various frames with rotation, translation and small scaling could be improved by a careful choice of the process condition and KLT pyramid implementation.

• Journal of Korean Society for Geospatial Information Science
• /
• v.18 no.2
• /
• pp.93-97
• /
• 2010

This paper presents the improved KLT(Kanade-Lucas-Tomasi) of registration of Image sequence captured by camera mounted on unmanned helicopter assuming without camera attitude information. It consists of following procedures for the proposed image registration. The initial interested points are detected by characteristic curve matching via dynamic programming which has been used for detecting and tracking corner points thorough image sequence. Outliers of tracked points are then removed by using Random Sample And Consensus(RANSAC) robust estimation and all remained corner points are classified as inliers by homography algorithm. The rectified images are then resampled by bilinear interpolation. Experiment shows that our method can make the suitable registration of image sequence with large motion.

• Smart Structures and Systems
• /
• v.31 no.4
• /
• pp.409-419
• /
• 2023

Structural integrity can be accessed from dynamic deformations of structures. Moreover, dynamic deformations can be acquired from non-contact sensors such as video cameras. Kanade-Lucas-Tomasi (KLT) algorithm is one of the commonly used methods for motion tracking. However, averaging throughout the extracted features would induce bias in the measurement. In addition, pixel-wise measurements can be converted to physical units through camera intrinsic. Still, the depth information is unreachable without prior knowledge of the space information. The assigned homogeneous coordinates would then mismatch manually selected feature points, resulting in measurement errors during coordinate transformation. In this study, a two-stage optimization method for video-based measurements is proposed. The manually selected feature points are first optimized by minimizing the errors compared with the homogeneous coordinate. Then, the optimized points are utilized for the KLT algorithm to extract displacements through inverse projection. Two additional criteria are employed to eliminate outliers from KLT, resulting in more reliable displacement responses. The second-stage optimization subsequently fine-tunes the geometry of the selected coordinates. The optimization process also considers the number of interpolation points at different depths of an image to reduce the effect of out-of-plane motions. As a result, the proposed method is numerically investigated by using a truss bridge as a physics-based graphic model (PBGM) to extract high-accuracy displacements from recorded videos under various capturing angles and structural conditions.

• Journal of the Korean Society for Aviation and Aeronautics
• /
• v.28 no.4
• /
• pp.1-11
• /
• 2020

In order to diagnose low-level turbulence in Korea, diagnostic indices of low-level turbulence were calculated from Aug 2016 to Jul 2019 using a Korea Meteorological Administration Post Precessing (KMAPP) developed by the National Institute Meteorological Sciences (NIMS), and the indices were evaluated using Aircaft Meteorological Data Relay (AMDAR). In the mean horizontal distribution of diagnostic indices calculated, severe turbulence was simulated along major domestic mountains, including near the Taebaek Mountains, the Sobaek Mountains and Hallasan Mountain on Jeju Island due to geographical factors. Later, detection performance was evaluated by calculating the KMAPP Low-Level Turbulencd index (KLT) on combined index, using AUC value of Individual diagnostic indices as a weight. The result showed that the AUC value of KLT was 0.73, and the detection performance was improved (0.02-0.13) when the index was combined. Also, when looking for the AMDAR data is divided into years, seasons, and altitudes, up to 0.94 AUC values were found in winter (DJF) and the surface (surface-1,000ft). By using high-resolution numerical data reflecting detailed terrain data, local turbulence distribution was well demonstrated and high detection performance was shown at low-level.

• Journal of the Korea Society of Computer and Information
• /
• v.14 no.3
• /
• pp.65-74
• /
• 2009

Scene change detection is pretreatment to index and search video information in video search system, and it is very important technology for overall performance. Existing scene change detection used single characteristic of pixel value difference, histogram difference, etc or mixed single characteristics that have complementary relationship. However, accuracy of those researches is very poor for special video such as infrared camera, night shooting. Therefore, this paper is proposed the method that is mixed color histogram and at algorithm for scene change detection at the specific movie. To verify the usefulness of a proposed method, we did an experiment which used color histogram only and KLT algorithm with color histogram. In result, evaluation index of proposed method is improved about 11.4% at the specific movie.

• The Journal of the Acoustical Society of Korea
• /
• v.20 no.4
• /
• pp.35-41
• /
• 2001

This paper describes Linear Discriminant Analysis and common vector extraction for speech recognition. Voice signal contains psychological and physiological properties of the speaker as well as dialect differences, acoustical environment effects, and phase differences. For these reasons, the same word spelled out by different speakers can be very different heard. This property of speech signal make it very difficult to extract common properties in the same speech class (word or phoneme). Linear algebra method like BT (Karhunen-Loeve Transformation) is generally used for common properties extraction In the speech signals, but common vector extraction which is suggested by M. Bilginer et at. is used in this paper. The method of M. Bilginer et al. extracts the optimized common vector from the speech signals used for training. And it has 100％ recognition accuracy in the trained data which is used for common vector extraction. In spite of these characteristics, the method has some drawback-we cannot use numbers of speech signal for training and the discriminant information among common vectors is not defined. This paper suggests advanced method which can reduce error rate by maximizing the discriminant information among common vectors. And novel method to normalize the size of common vector also added. The result shows improved performance of algorithm and better recognition accuracy of 2％ than conventional method.